Letterly
Letterly pairs students with live tutors and needed AI that helps the tutor without touching the child's work. We designed and built the product and support it with a scaling team.
- Sector
- EdTech · literacy
- Services
- UX · AI Enablement · Development · Managed Service
- Model
- Scaling team
- Status
- Live · ongoing support
What they do
Letterly pairs students with live private tutors, working to individual goals through structured bootcamps that fit around school. Students build a portfolio of published work, with pieces appearing in The Journal, and progress is tracked against writing benchmarks with daily feedback.
Their published outcome: students improve literacy by half a grade level in eight weeks, around twice the rate of school.
What it was costing them
AI, children and writing is the most contested combination in education. Schools are hostile to anything that looks like it writes for the student, with good reason.
The product only has value if the student genuinely did the writing, so any feature that blurs that destroys what is being sold.
Children are users here, which brings the UK Age Appropriate Design Code into scope for data, defaults and nudges.
The work
We pointed the AI at the tutor's workload rather than the student's work: structuring feedback, tracking progress against benchmarks, and surfacing patterns across a portfolio that a tutor seeing one piece a week would miss.
The student side of the product stays deliberately quiet. The AI does not generate student work and does not grade in place of a human.
We support the platform with a scaling team, which is how Letterly grew delivery without growing headcount at the same rate.
The line schools want stated
Every school leader is drawing the same line, and very few suppliers will say where theirs is.
Ours: the AI can observe, structure and summarise a student's work. It cannot produce it and it is not the final word on its quality. A human tutor sits between the model and the child.
Saying that plainly is what gets the product through a school's own review.
What changed
- ½ grade / 8 wks
- Client's published literacy outcome
- 56%
- Delivery efficiency via scaling team
The 56% figure needs a stated baseline and measurement period before it carries real weight — we are agreeing that with the client.
Similar work elsewhere
3
Membership audiences served
Propolis
A membership community three audiences stay in, launched in 2021 and still growing.
3-level
DfE standard mapped end to end
Barkers EPM
School trusts can evidence their estate against the DfE standard without an estates team.
Scalewise
Could this work for your business?
Thirty minutes with the people who would do the work. You will leave with a view on whether it is worth building and roughly what it costs.